ENHANCE Enhancing Risk Management Partnerships for Catastrophic Natural Disasters in Europe Grant Agreement number 308438 Deliverable 7.3: RISK SCENARIOS AND ANALYSIS LONDON CASE STUDY: FLOOD RISK AND CLIMATE CHANGE IMPLICATIONS FOR MSPs
ENHANCE Enhancing Risk Management Partnerships
for Catastrophic Natural Disasters in Europe
Grant Agreement number 308438
Deliverable 7.3: RISK SCENARIOS AND ANALYSIS
LONDON CASE STUDY: FLOOD RISK AND CLIMATE CHANGE
IMPLICATIONS FOR MSPs
Project 308438 • Risk scenarios and analysis: London case study ii
Title
RISK SCENARIOS AND ANALYSIS: LONDON CASE STUDY: FLOOD RISK
AND CLIMATE CHANGE IMPLICATIONS FOR MSPs
Author(s) Katie Jenkins, Jim Hall (UOXF); Swenja Surminski, Florence Crick (LSE)
Organization
UOXF, LSE
Deliverable Number
D 7.3
Submission date
26-02-2015
Prepared under contract from the European Commission Grant Agreement no. 308438 This publications reflects only the author’s views and that the European Union is not liable for any use that may be made of the information contained therein. Start of the project: 01/12/2012 Duration: 48 months Project coordinator organisation: IVM Due date of deliverable: Month 27 Actual submission date: Month 27 Dissemination level
PU Public
PP Restricted to other programme participants (including the Commission Services)
RE Restricted to a group specified by the consortium (including the Commission Services)
CO Confidential, only for members of the consortium (including the Commission Services)
Project 308438 • Risk scenarios and analysis: London case study iii
Executive summary
Flooding is recognised as one of the most common and costliest natural disasters in England and is
listed as a major risk on England's National Risk Register. Floods can take various forms, such as
river, coastal and surface water flooding. Surface water flooding represents flooding in urban areas
during heavy rainfall due to a combination of factors, including pluvial flooding (resulting from heavy
rainfall which does not infiltrate the ground but ponds or flows overland) flooding from sewers,
drains, and small watercourses. In the UK property is more likely to experience repeated surface water
flooding compared to fluvial or coastal flooding, costing England an estimated £1.3bn to £2.2bn per
year (Defra, 2011). In London surface water flooding is considered to be the most likely cause of
flood events, and probably the greatest short-term climate risk (Greater London Authority,
2009, 2011).
Additional to this threat is the potential impact of climate change on the frequency and intensity of
heavy precipitation events (IPCC, 2013). The UK Climate Projections (UKCP09) revealed that over
the course of this century, rainfall is projected to increase in winter but decrease in the summer,
although the number of days of heavy rainfall in summer will increase. Summers are therefore likely
to be characterised by intense heavy rainfall events intermixed with longer and relatively drier periods.
Winters, on the other hand will become wetter, with not only more average rainfall but also an
increase in extreme winter precipitation. These changes in winter and summer rainfall are expected to
lead to an increase in fluvial and surface water flooding (Ramsbottom et al, 2012).
In Greater London the number of residential properties in areas prone to surface water flooding has
been increasing from 2001 to 2011, as has the proportion of urban land covered with manmade
surfaces (which is >70% in many London boroughs) (HR Wallingford, 2012). Development in areas
prone to surface water flooding has been estimated at 0.5 to 0.7% per year from 2008 (Adaptation
Sub-Committee, 2012). These issues highlight that current land-use and development plans may
increase the exposure and vulnerability of society to surface water flood risk in Greater London.
Consequently, Defra estimate flood damage from surface water run-off could increase by 60-220%
over the next 50 years linked to changing precipitation patterns due to climate change, and
urbanisation (Adaptation Sub-Committee, 2012). The increased risk of surface water flooding and
need for further information has been highlighted as an area of concern within the London Plan (GLA,
2011), with effective and economically viable adaptation options required.
A particularly interesting aspect of flood management in England is the public-private partnership on
flood insurance between the UK government and the insurance industry known as the Statement of
Principles. Flood insurance is underwritten by the private market, while government commits to flood
risk management activities. However, the 2007 UK floods triggered a review of the Statement of
Principles. After more than two years of negotiation between government and industry, a new flood
insurance system Flood Re, was proposed by government in summer 2013, with the aim to finalise
and implement the new scheme by mid-2015. The proposed system, which creates an insurance pool
for properties at high risk of flooding, is presented by government and industry as a roadmap to future
affordability and availability of flood insurance, with an anticipated run-time of 20 to 25 years (Defra
and ABI 2013). The mechanisms of the new Flood Re scheme are still being negotiated, and to date
Project 308438 • Risk scenarios and analysis: London case study iv
there is little mention of how the MSP between government and insurers, and the new Flood Re
scheme could also promote effective flood risk reduction measures.
The London case study aims to provide an investigation of this public-private partnership and the
specific issue of surface water flood risk, and aims to analyse how the current and future proposals for
the partnership could influence London’s resilience to surface water flood risk today and in the future.
A key consideration is the incentives for risk reduction among different partners (including the
government, insurers, and developers), to support flood defences, household level flood protection,
and more appropriate spatial planning and zoning.
Crucial input into this investigation is a surface water flood risk assessment for Greater London. The
report presents a framework for evaluating current and future risk of surface water flooding in Greater
London. The study uses probabilistic hourly rainfall data from a spatially coherent version of the
UKCP09 Weather Generator (WG) (Kilsby et al., 2011) to identify heavy precipitation events for a
range of climate scenarios. A link is established to detailed Drain London1 surface water flood depth
maps, and damage to residential properties estimated using depth-damage functions.
The flood risk analysis highlights how this risk is expected to change in the future under projections of
climate change. Daily event frequency could increase by 54% by the 2030s and by 60% by the 2050s
compared to the baseline period. Average annual damage to residential properties is projected to
increase by 50% and 80% by the 2030s and 2050s respectively, suggesting an increase in severity of
events when they do occur. Based on the daily event data, the expected annual damage (EAD) is
calculated as £103, £184 and £198 million/year for the baseline, 2030 high and 2050 high climate
change scenarios respectively. An important component of the spatial WG is also the ability to provide
probabilistic results, with the range in results increasing from the baseline period to the future time
periods, representing the climate model uncertainty underlying the precipitation projections.
As a second step the event and damage data set is used as input within an agent-based model (ABM).
The model is parameterised based on large array of data sources and developed around GIS data to
allow a realistic representation of residential buildings and surface water flood risk. The ABM is
developed such that it can demonstrate the effect of flood risk and insurance on household wealth and
the potential for spatial shifts in inequality as a consequence of flood damage and insurance
(un)availability; assess the role of flood defences and PLPMs for risk reduction; and investigate the
existing public-private flood insurance partnership and the proposed new insurance scheme Flood Re.
The risk analysis provides a useful tool for decision making and risk management, and through the
proposed incorporation of this data in the ABM aims to support and justify future adaptation
strategies. These will be particularly important for Greater London where surface water flooding is
considered to be the most likely cause of flood events, and probably the greatest short-term climate
risk.
This will also be of great benefit to the MSP to highlight the potential benefits and limitations of the
scheme for risk reduction under future scenarios of climate change, broader economic consequences
for households and the housing market, potential benefits of investment in flood defences and
1 Led by the Greater London Authority. See: http://www.london.gov.uk/drain-london
Project 308438 • Risk scenarios and analysis: London case study v
household level protection in terms of reduced damages and reduced financial risk for insurers; and
the potential issues of future development in high risk areas for the scheme. Other components of the
scheme can also be tested, such as the inclusion of certain types of properties in Flood Re and the
financial implications of different transitional pathways to risk-based pricing of insurance in the
longer-term.
The above issues and questions are all highly relevant aspects for the ongoing regulatory and political
approval process for Flood Re, which have until now not received sufficient attention due to lack of
data or analysis. Our work addresses this gap and our findings are expected to provide important input
to the current discussion about the design and operation of Flood Re, particularly with regards to
incentivising flood risk reduction measures.
Project 308438 • Risk scenarios and analysis: London case study vi
Contents
1 Introduction ....................................................................................................................... 1
1.1 Surface water flood risk ........................................................................................................ 1
1.2 Flood risk management ......................................................................................................... 1
1.3 Aims and Objectives .............................................................................................................. 2
2 Specification of the risk analysis ...................................................................................... 4
2.1 Surface water flood risk in London ..................................................................................... 4
2.2 Modelling Framework........................................................................................................... 6
2.3 Spatial weather generator for urban areas ......................................................................... 7
2.4 Flood damage estimates ........................................................................................................ 8
2.5 Flood risk, insurance and adaptation strategies: An agent based model approach ........ 9 2.5.1 Modelling flood risk within the ABM ........................................................................... 12
2.5.2 Incorporating adaptation options ................................................................................... 12
2.5.3 Modelling the insurance market and Flood Re scheme ................................................. 14
2.5.4 Modelling the London housing market ......................................................................... 15
2.5.5 Modelling future development ...................................................................................... 16
2.5.6 Using the ABM to inform MSPs ................................................................................... 17
3 Results .............................................................................................................................. 18
3.1 Surface water flood risk analysis ....................................................................................... 18
4 Discussion of Results ....................................................................................................... 27
4.1 Surface water flood risk analysis ....................................................................................... 27
5 Conclusions ...................................................................................................................... 30
6 References ........................................................................................................................ 32
Project 308438 • Risk scenarios and analysis: London case study 1
1 Introduction
1.1 Surface water flood risk
Flooding is recognised as one of the most common and costliest natural disasters in England and is
listed as a major risk on England's National Risk Register. Floods can take various forms, such as
river, coastal, surface water, sewer and groundwater flooding, occurring independently but also in
conjunction, resulting in wide reaching impacts and consequences. Flooding affects not only people
and communities but buildings, agricultural land, infrastructure, ecosystems and often results in severe
displacement after an event. Flood risk management has been identified by the UK Climate Change
Risk Assessment as one of the high priority areas for action for the UK over the next five years, with
socio-economic factors including population growth, a changing climate, new development in the
floodplain and the relatively little understood surface water risk currently seen as the main challenges.
According to the UK CCRA coastal and surface flooding is the biggest potential impact of climate
change on the UK with the prospect for increasing losses and damage.
When comparing literature on fluvial or coastal flood risk, and surface water flood risk the latter is less
commonly investigated, and published articles specifically dealing with this type of risk are less
common. Correspondingly, there has been much less emphasis on conducting quantitative assessments
of economic and social costs of surface water flooding under future climate change, despite the
potential serious nature of impacts. At a broader level concerns have been raised that pluvial flood risk
across Europe may not be fully recognised, with some member states giving a much lower priority to
this type of risk (European Water Association, 2009).
In the UK property is more likely to experience repeated pluvial flooding compared to fluvial or
coastal flooding (Dawson et al., 2008), costing England an estimated £1.3bn to £2.2bn per year (Defra,
2011). Surface water flooding represents flooding in urban areas during heavy rainfall due to a
combination of factors, including pluvial flooding2, flooding from sewers, drains, and small
watercourses (European Water Association, 2009; Falconer et al., 2009). As such, heavy rainfall is
only one component in determining surface water flood risk. Drainage, often linked to sewage systems
in urban areas, and infiltration capacity related to land-use are also important factors. In the UK the
majority of surface water runoff drains directly into sewers, even from new developments.
Furthermore, as surface water flooding is often characterised by short duration and high intensity
rainfall events and so is difficult to predict, hindering the ability to forecast and warn communities of
flood risk.
1.2 Flood risk management
Flood management responsibility, policy and legislation for England are determined by the
Department for Environment, Food and Rural Affairs with national flood and coastal erosion
2 Pluvial flooding can be defined as flooding resulting from heavy rainfall which does not infiltrate the ground but ponds or
flows overland before the runoff enters a natural or man-made drainage system or watercourse, or where water cannot enter a
system as it is already at full capacity. Pluvial flooding is often characterised by short duration high intensity rainfall events
(typically over 20mm/h) (European Water Association, 2009).
Project 308438 • Risk scenarios and analysis: London case study 2
management delivered by the Environment Agency. Local authorities have lead responsibility for
managing local flood risk, which includes surface water runoff, groundwater and ordinary
watercourses, and are designated as Lead Local Flood Authorities.
A particularly interesting aspect of flood management in England is the public-private partnership on
flood insurance between the UK government and the insurance industry known as the Statement of
Principles. Flood insurance across the United Kingdom is unique amongst most other national
schemes as it is purely underwritten by the private market, while government commits to flood risk
management activities. The Statement of Principles was established in 2000 in the wake of growing
flood losses and sets commitments from both the insurance industry and government to establish flood
insurance provision. The main obligations can be summarised as follows: flood insurance is provided
by private insurers under the Statement of Principles to both households and small businesses,
generally up to a risk level of 1:75 return period (RP) (1.3%) as part of their building and/or content
cover. Properties at higher risk are granted cover if insurers are informed by the Environment Agency
about plans for flood defence improvements for the particular area within the next five years.
Government commits to investment in flood defences and improved flood risk data provision as well
as a strengthened planning system. Under this agreement, the emphasis on flood risk reduction is
primarily placed on the government (national and local) as insurers play more of a financial supporting
role with little mention of how insurance can promote effective flood risk reduction measures.
The 2007 UK floods triggered a review of the Statement of Principles. After more than two years of
negotiation between government and industry, a new flood insurance system Flood Re, was proposed
by government in summer 2013. The Statement of Principles (SoP) officially ended on the 30th June
2013, but is still in operation whilst the political debate about the new Flood Re system continues,
with the aim to finalise and implement the new scheme by mid-2015. The proposed system, which
creates an insurance pool for properties at high risk of flooding, is presented by government and
industry as a roadmap to future affordability and availability of flood insurance, with an anticipated
run-time of 20 to 25 years (Defra and ABI 2013).
The Flood Re scheme will provide households under low to normal risk with standard insurance and
high risk properties will be insured through the Flood Re pool. The subsidy for high risk households is
claimed from a levy taken from all policyholders, on average £10.50 per policy, and also imposed on
insurers according to their market share. The premiums offered for high risk households are fixed
dependent on council tax banding and cover is offered at a set price based on what is felt to be initially
affordable. The government proposal is that small businesses will not be covered by the Pool unless
they operate from home with a domestic insurance policy in place. Policy excesses are intended to be
limited to between £250 and £500. Several other technical aspects remain unclear, including the
handling of flood losses beyond a suggested cap of 1 in 200 loss event, and will be subject to debate
between insurers and government.
1.3 Aims and Objectives
The main focus of the London case study is an investigation of the public-private partnership on flood
insurance between the UK government and the insurance industry. The case study aims to analyse
Project 308438 • Risk scenarios and analysis: London case study 3
how this partnership could influence London’s resilience to surface water flood risk today and in the
future. A key consideration is the incentives for risk reduction among different partners, to support
flood defences, household level flood protection, and spatial planning and zoning, and how the
existing and proposed insurance schemes address this.
To achieve the objectives of the case study the analysis is conducted in two main parts:
i. The study uses probabilistic hourly rainfall data from a spatially coherent version of the
UKCP09 Weather Generator (WG) (Kilsby et al., 2011) to identify heavy precipitation events
for a range of climate scenarios. A link is established to detailed Drain London3 surface water
flood depth maps, and damage to residential properties estimated using depth-damage
functions. This facilitates the quantification of surface water flood risk in Greater London.
ii. Outputs from (i) are used within an agent-based model (ABM) developed for London. The
model is parameterised based on large array of data sources and developed around GIS data to
allow a realistic representation of residential buildings and surface water flood risk. The ABM
is developed such that it can demonstrate the effect of flood risk and insurance on household
wealth and the potential for spatial shifts in inequality as a consequence of flood damage and
insurance (un)availability; assess the role of flood defences and PLPMs for risk reduction; and
investigate the existing public-private flood insurance partnership and the proposed new
insurance scheme Flood Re.
The main focus of this report is related to part (i), outlining the analysis of future flood risk and
results. However, an introduction and overview of the ABM, its application to flood risk, adaptation
scenarios, and how it will be used to inform MSPs, is also provided.
3 Led by the Greater London Authority. See: http://www.london.gov.uk/drain-london
Project 308438 • Risk scenarios and analysis: London case study 4
2 Specification of the risk analysis
2.1 Surface water flood risk in London
The study focuses upon Greater London (Figure 1), encompassing an area of 1,574 km2, a population
of approximately 7.5 million people, and 3.3 million residential dwellings (ONS, 2001).
Figure 1: The boundary of Greater London study area and location in England (inset)
London is considered well protected against tidal flooding but has a relatively low standard of
protection against surface water flooding. For example, surface water flooding in London in July 2007
affected over 400 properties, 158 schools and two hospitals (Environment Agency, 2010).
Consequently, in London surface water flooding is considered to be the most likely cause of flood
events, and probably the greatest short-term climate risk (Greater London Authority,
2009, 2011). More than 800,000 properties have been estimated to be at risk, and while most drainage
systems are designed to cope with a 1/30 year storm event, maintenance is often poor and so parts of
the network can perform below these standards (ibid.).
In Greater London the number of residential properties in areas prone to surface water flooding has
been increasing from 2001 to 2011, as has the proportion of urban land covered with manmade
surfaces (which is >70% in many London boroughs) (HR Wallingford, 2012). Over 96% of new
developments in London have been on brownfield sites, however, many of the remaining brownfield
sites for development lay in flood risk zones, with limited alternative sites for large scale development
(Greater London Authority, 2009). Similarly, development in areas prone to surface water flooding
has been estimated at 0.5 to 0.7% per year from 2008 (Adaptation Sub-Committee, 2012). These
Project 308438 • Risk scenarios and analysis: London case study 5
issues highlight that current land-use and development plans may increase the exposure and
vulnerability of society to surface water flood risk in Greater London.
Additional to this threat is the potential impact of climate change on the frequency and intensity of
heavy precipitation events (IPCC, 2013). The UK Climate Projections published in 2009 (UKCP09)
revealed that over the course of this century, rainfall is projected to increase in winter but decrease in
the summer, although the number of days of heavy rainfall in summer will increase. Summers are
therefore likely to be characterised by intense heavy rainfall events intermixed with longer and
relatively drier periods. Winters, on the other hand will become wetter, with not only more average
rainfall but also an increase in extreme winter precipitation which is expected across the UK in all
regions. Projections for the 2080s under a medium emissions scenario suggest that in winter increases
in rainfall will vary between +10% and +30% over most of the UK, while in the summer there appears
to be a south to north gradient with a decrease in rainfall of close to 40% in south-west England to
almost no change in the north of Scotland ((Murphy et al., 2009)). These changes in winter and
summer rainfall are expected to lead to an increase in fluvial and surface water flooding ((Ramsbottom
et al., 2012)). Similarly, in England, Maraun et al., (2008) identified long-term increases in the
intensity of winter precipitation, and more recently increasing trends for spring and autumn. Fowler
and Ekström (2009) also projected increasing precipitation extremes in the UK for spring, autumn and
winter, increasing by 5-30% across different regions and seasons by 2070–2100.
Consequently, Defra estimate flood damage from surface water run-off could increase by 60-220%
over the next 50 years linked to changing precipitation patterns due to climate change, and
urbanisation (Adaptation Sub-Committee, 2012). The increased risk of surface water flooding and
need for further information has been highlighted as an area of concern within the London Plan (GLA,
2011), with effective and economically viable adaptation options required.
The Flood and Water Management Act 2010 gives London boroughs clearer responsibilities related to
surface water flood risk. New developments should utilise sustainable urban drainage systems (SUDS)
where possible, which aim to manage rainwater falling on roofs and other surfaces through a sequence
of actions. The main objective is to manage flow rate and volume of surface water runoff to reduce the
risk of flooding. Yet, the current uptake of SUDS is insufficient to mitigate increasing flood risk from
surface runoff, the risk of sewer overload, or to protect water quality (Defra, 2011). Incorporating
SUDS into new developments will only be part of the solution as new developments account for a
very small fraction of the housing stock. As such, retrofitting SUDS within existing urban areas and
developments will also be important for managing future flood risk (Environment Agency, 2007).
SUDS can also alleviate pressure on the sewerage network which will be crucial as in the UK the
majority of surface water runoff drains directly into sewers. Ellis and Viavattene (2014) estimate that
at least two-thirds of current urban flooding in the UK is the result of drainage system failure. Most
drainage systems in the UK are designed to cope with a 1/30 year storm event, but 50% of sewerage
systems are currently at or beyond capacity demanding major investment of an estimated £600m per
year (Defra, 2011).
Project 308438 • Risk scenarios and analysis: London case study 6
Measures can also be implemented at a private level to reduce the severity of flood damage. For
example, resistance measures to keep water out such as door barriers, and resilience measure to reduce
the depth and damage caused by flood water like the use of water pumps and waterproof plaster. The
costs of PLPMs are reported to range from £3000 to £10,000 (Defra, 2008b; Harries, 2012), and have
been estimated to reduce flood damage costs by 65-84% (Thurston et al., 2008).
2.2 Modelling Framework
Risk based information on the potential economic and social impacts of surface water flooding, and
assessments of the effectiveness of adaptation measures, will be highly relevant if policy makers are to
address such concerns in the longer-term, and strive to implement more effective and efficient
adaptation measures. In developing such methodologies it is essential to recognise that surface water
flood risk and its associated impacts can be influenced by a variety of factors including the intensity
and spatial extent of the rainfall event itself; topography of the area; type of land use and infrastructure
in the region affected; permeability of the ground; surface and underground drainage capacity; and
interaction of pluvial flooding with other flood types such as fluvial flooding (European Water
Association, 2009). Furthermore, when focusing on complex and interrelated urban systems this
means that assessments of surface water flood risks and their impacts can span a wide range of
disciplines, government agencies, and actors.
Secondly, due to the characteristics of surface water flood events methodologies need to de developed
which can appropriately capture the localised scale and nature of the events, as well as the specific
characteristics of the urban areas affected. The scale and severity of economic and social impacts will
be dependent on the underlying vulnerability of the population and particular region exposed to the
event, as well as the underlying climate and weather patterns that determine the frequency and severity
of the event itself (Hall et al., 2005). This requires integrative thinking to understand and model
relationships between the urban environment, climate change, flood risk, and policy responses at a
local scale.
Fig. 2 and the below text provides an overview of the methodological framework (further details on
the main components are included in sections 2.3 – 2.4). The study uses flood depth maps for 1/30,
1/100, and 1/200 year return periods generated by the Drain London project. These maps reflect
present day climate, topographic data and information on the surface water drainage system at a fine
resolution (5 x 5m). Economic damages to residential building fabric and contents are calculated using
published flood depth-damage functions (Penning-Rowsell et al., 2010). To assess the potential impact
of climate change on future flood risk and economic damages, the corresponding return level of
extreme precipitation events of 1/30, 1/100, and 1/200 year return periods are estimated using hourly
rainfall data from a spatial version of the UKCP09 WG (Kilsby et al., 2011). This facilitates a
probabilistic analysis of extreme rainfall events, as well as providing an assessment of underlying
climate model uncertainties. Thus, for each flood event identified the spatial extent can be mapped, the
flood depth ascertained from the Drain London maps, and economic damage to residential buildings
estimated using depth-damage functions.
Project 308438 • Risk scenarios and analysis: London case study 7
Figure 2: Overview of the risk based modelling framework which feeds into the ABM
2.3 Spatial weather generator for urban areas
The spatial and temporal scale of climate model outputs is often inconsistent with that required for
climate change impact studies. More spatially explicit climate projections can be produced by
incorporating downscaling techniques that account for local climatological features. The most recent
UK climate scenarios (UKCP09) have been accompanied by a stochastic Weather Generator (WG)
which can provide daily and hourly time series of weather variables for present and future conditions
at a 5 km2 resolution (Jones et al. 2009). The WG incorporates a stochastic rainfall model, which
simulates future rainfall sequences, and then generates other weather variables according firstly to the
rainfall state and then to other inter-variable relationships which are represented as regression
relationships. The WG has been well validated against observed data from 1961 to 1990 (ibid.).
The UKCP09 scenarios were novel in their representation of climate model uncertainties, based on the
range of climate model responses from a large perturbed physics ensemble (Murphy et al. 2009).
Results are presented as probability distributions of projected changes which can be used to
parameterise changes in the WG. In practice this is achieved by providing a Monte Carlo sample of
10,000 equiprobable vectors of change factors that are used to parameterise the WG. This leads to a
two-level sampling scheme in which (i) repeated representations of the WG for a given vector of input
parameters can be used to explore the effects of natural variability and (ii) sampling different vectors
of change factors explores the effect of climate model uncertainty in projected future impacts. In the
results reported here we refer to climate projections explored via (ii).
UKCP09 Projections
Weather Generator
Climate
Scenarios
Return level
(mm/hr) for return
periods (baseline)
Flood Risk Analysis
Building fabric &
content damage
dataset
Drain London flood
depth maps (1/30,
1/100, 1/ 200 year
return periods)
Database of extreme
rainfall events
Residential building
database (build type
/ age / location)
Depth-Damage
Functions
Direct economic
costs of surface
water flood events
Project 308438 • Risk scenarios and analysis: London case study 8
The UKCP09 WG simulates weather sequences at a single site so does not provide spatial consistency
in time across neighbouring grid cells (Jones et al. 2009). The lack of spatial coherence limits the use
of the WG for analysing aggregate impacts over several grid cells. In this study a modified version of
the UKCP09 WG is used which provides spatially coherent time series data. Hourly precipitation
time-series data for 30 year stationary sequences (t) are taken from the WG for each grid cell (g) in the
study area. These series are generated 100 times each, with each run (r) based on a different randomly
sampled vector of change factors, to allow probabilistic analysis. This results in a 3-dimensional data
array of precipitation values A = [𝐴𝑔,𝑡,𝑛], where g = 71, t = 263,520, and r = 100. Data is generated for
the baseline period (1961–1990) and for the 2030s and 2050s under high emission scenarios
(equivalent to the IPCC SRES B1 and A1FI scenarios).
In order to assess surface water flood risk the return level of extreme precipitation events for 1/30,
1/100, and 1/200 year return periods is estimated from the baseline data. To calculate the recurrence
interval the hourly annual maximum series (AMS) is derived from the 100*30 year precipitation
times-series data for each grid cell. Extreme Value Analysis (EVA) is used to calculate the return
levels for each return period. The Generalised Extreme Value (GEV) distribution function is fitted to
the AMS. EVA allows the probability and return levels of extreme events to be determined, for return
periods exceeding that of the original data series, and even if events are more extreme than exists in
the data series (Sanderson, 2010). For each return period the GEV distribution allows the equivalent
return level to be determined for each grid cell. This gridded data provides the precipitation thresholds
above which surface water flooding of a given return period would be assumed to occur in each grid
cell. This is linked to damage costs of affected residential properties which intersect each grid cell in
the study area (outlined in section 2.4 below). These look up tables allow damages to be estimated for
each spatially explicit surface water flood event identified for the baseline and the 2030 and 2050 high
emission scenarios.
2.4 Flood damage estimates
Drain London provide surface water flood depth maps for Greater London based on precipitation data,
topographic data and information on the surface water drainage system. Flood risk is considered to
occur where flood depths exceed 0.1m as this is the depth where the significant onset of impacts, e.g.
damage to property, is considered to transpire (Penning-Rowsell et al., 2010). This study uses the UK
Buildings residential building class dataset4 which provides the spatial footprint of properties, property
type, and age for Greater London (It is assumed that there is no change in the number or spatial pattern
of residential buildings when assessing future scenarios). By overlaying the spatial flood maps onto
the building data it is possible to identify which properties are at risk of surface water flooding, and
the flood depth. As is the standard approach for flood loss analysis in the UK this study utilises the
damage functions provided in the Multi-Coloured Manual (MCM), for short (<12hr) duration floods
(Penning-Rowsell et al., 2010) (for more details on the application of depth damage functions also see
inter alia. Messner et al., (2007). Damage to residential building fabric and contents are estimated
4 The GeoInformation group data ® copyright by The GeoInformation® Group, 2014 Licence No. 3786.
Project 308438 • Risk scenarios and analysis: London case study 9
based on building type (detached; semi-detached; terraced; flat), and building age (unknown; pre-
1919; 1919-1944; 1945-64; 1965-79; 1980-current).
Surface water flooding would not affect the whole of Greater London at the same moment in time. A
main benefit of the WG is that it is spatially coherent so an event affecting multiple grid cells can be
assessed in terms of damage for each property type and scenario, and results mapped to show the
spatial pattern of damages. In addition to assessing the economic damages from individual surface
water flood events, the average annual damage (EAD) can also be estimated based on the product of
the probability of a given flood event and damage integrated across all the scenario runs (e.g.
following Hall et al., (2006).
In order to summarise the flood event data and provide clear results, a daily event is defined as any
day when surface water flooding occurs in one or more grid cells for 1 hour or more. In the cases
where the same grid cell is affected for 2 or more consecutive hours the maximum hourly damage is
used in the estimates.
In addition, the spatial time series of flood events, and associated damages to residential properties,
provides the event set on which the effects of current and proposed public-private flood insurance
partnerships and incentives for adaptation will be tested within the ABM. In parallel with the
assessment of flood insurance, incentives for property level protection measures and flood defences
are also represented within the ABM.
2.5 Flood risk, insurance and adaptation strategies: An agent based model approach
ABMs are useful as they provide a bottom-up approach for understanding systems and their
behaviour, and are advantageous for visualising the effects of changing behaviours. To date ABM has
had limited application specifically to the insurance sector. Two papers have been identified which
look at consumer behaviour and insurance choices in general (Ulbinaitė et al., 2011; Ulbinaitė and Le
Moullec, 2010), while Brouwers and Boman (2011) focus more specifically on flood management
strategies, economic consequences, and decision making regarding flood insurance. More recently an
ABM has been developed which combines sea level rise and the impact of a flooding insurance
program, and models how these two aspects are likely to affect households’ location choices in a
coastal town in the US (Putra et al., 2014 submitted; Zhang et al., 2013). These studies establish a
bottom-up analytical framework to show how local real-estate markets respond to flood events and
how local governments choose adaptive actions.
In developing the ABM model code is adapted from the Putra (2014) model to capture the effects of
flood events and insurance on the local housing market. This code is adapted to the specific
characteristics of the UK and London, and extended to model insurance in a more elaborate fashion.
The ABM is parameterised based on large array of data sources and developed around GIS data to
allow a realistic representation of residential buildings and surface water flood risk. Fig. 3 provides an
overview of the model structure and design.
Project 308438 • Risk scenarios and analysis: London case study 10
Figure 3: Overview of the agent based model
The ABM characterises five different agents: property owners, insurers, developers, government, and
bank agents, who interact within the environment. The key actions/features of the agents are
summarised as:
Property owners
Classed in the model as homeowners, buyers or sellers.
Decisions to sell are based on ability to pay mortgage and house fees, and the value of
property
Allocated incomes, housing fees, insurance costs
Homeowners are assumed to have flood insurance
People can also invest in Property Level Protection Measures (PLPMs)
- Classed as proactive and reactive responders to flood risk
Insurer
Main task to provide insurance to people
Sets premium and excess for insurance, based on risk
Allocate households to Flood Re scheme
Government
Can build/increase flood defences based on CBA
Give grants to people investing in PLPMs
The local government can sell land and evaluate development plans
Collect taxes from property owners and land sales, and grant from central government for
flood defences
Developer
React to the housing demand in the model
Locate land for building within opportunity areas
Project 308438 • Risk scenarios and analysis: London case study 11
Submits development proposal to be approved by the local government.
Builds new houses and sells these houses on the market
Focus on making profit
Bank
A basic representation to allow the repossession of houses and selling these again on the
market
The properties and processes of these agents give rise to the aggregated results. The ABM runs in
annual time-steps and provides the user with options for spatial visualisation of key features (e.g.
house or person attributes (Figure 4)), switches to turn on/off different policies, flood events and
return periods, and graphs showing time-series of results. The following sections provide a description
of the main components, and agent behaviour, captured within the model.
Figure 4: Example of section of the user interface and visualisation: house view (build type)
Project 308438 • Risk scenarios and analysis: London case study 12
2.5.1 Modelling flood risk within the ABM
Following the initial set up of the ABM the next key step is to impose a flood event on the study area.
Two approaches to modelling the flood ‘shock’ can be considered. Firstly, the model allows the user
to impose a flood event/s at set point/s in time via the user interface. This reads data from the daily
event dataset (from part (i) of the analysis) which provides information on the houses affected and the
damage to property fabric and contents for a given return period. Focusing on a few single events or
successive events is useful for model testing and validation. However, as this would represent one
single random event affecting the chosen area at any given time it is less robust for drawing policy
conclusions.
A second approach and area of planned future research is to use the probabilistic time-series data for
the analysis to test different policy options for the present day climate, and also considering future
effects of climate change. The daily event time series data is aggregated to annual steps (see Figure 6
below for an example). Where more than one daily event occurs in a single year the most damaging
event is used.
As well as incorporating data on the spatial pattern and level of damage for a given flood event an
estimate of flood risk is also required. This is calculated endogenously within the ABM to capture
dynamic changes in the model, and to allow risk to be assigned to new residential buildings created by
the model. Flood risk is calculated based on Bevan and Hall (2014, p.17). In any given year (t), the
risk (ri,t) is given by:
𝑟𝑖,𝑡 = ∫ 𝐷(𝑥𝑡)𝑓(𝑥𝑡)𝑑𝑥𝑡
∞
0
Where, 𝐷(𝑥𝑡) is a damage function with 𝑥 changing overtime, 𝑓(𝑥𝑡) is the flood probability
distribution. The probability of houses being affected by surface water flooding (for 1 in 30, 1 in 100,
and 1 in 200 year events, and damage data is available from part (i) of the analysis. As only three
points are available assumptions are made to extrapolating out from these points to create a damage
probability graph for each house. Namely, it is assumed that the function is linear between the three
known points. As the damage probability function will never have a probability of zero the function is
forced to meet the axis by assuming that the damage with zero probability is the maximum damage
that can be done to a house. There are many factors which would influence this at a household level,
and given a lack of data this is currently assumed to be 20% of the building value. The slope of the
function is assumed to extend horizontally until it crosses the x axis (damage). A maximum limit is
put on this assuming that it is highly unlikely that a house will be hit by a flood more than once every
2 years (i.e. 50% probability). Flood risk is then calculated as the area under the line.
2.5.2 Incorporating adaptation options
A key focus of the ABM is the investigation of the public-private partnership on flood insurance
between the UK government and the insurance industry, and how this partnership could influence
London’s resilience to major flood risk. This includes incentivising flood defences and household
level protection and as such options to adapt to flood risk are incorporated within the ABM. Alongside
Project 308438 • Risk scenarios and analysis: London case study 13
the role of insurance and the Flood Re scheme PLPMs and SUDs, and government spending on flood
defences are incorporated. In the first case these options will be switched on and off in the model to
test them and their relative costs and benefits in isolation and combination (Table 1).
In the model it is assumed that people can be proactive and reactive in their behaviour to investing in
PLPMs. Harries (2008) reports that in 2004/05 6% of un-flooded households and 39% of previously
flooded households had taken steps to increase their resilience to flooding, and by 2008 the equivalent
figures remained almost unchanged at 9% and 34%, respectively. The value of 34% is used to reflect
the percentage of flooded households who will invest reactively in PLPMs in the model. However,
given that the flood protection association reported that in 2008 less than 5,000 homes had to date
taken approved measures (Defra, 2008a) a lower figure of 1% is used to represent the proportion of
proactive households in the model. The costs of PLPMs are estimated to range from £3000 - £10,000
(Defra, 2008b; Harries, 2012), with PLPLMs estimated to reduce flood damage costs by 65-84%
(Thurston et al., 2008).
Table 1: Scenarios to be tested in the ABM, including adaptation options
Current
insurance
system
Flood Re
system
Investment in
PLPMs
Investment in
flood defences
Off Off Off Off
ON Off Off Off
ON ON Off Off
Off Off ON Off
Off Off Off ON
ON ON ON Off
ON ON Off ON
ON ON ON ON
Within the ABM the local Government agent can approve or refuse applications for future
developments, and use their allocated flood protection budget to provide grants to houses to
incentivise PLPMs and build defences. A simple representation of the process of approving and
building flood defences is captured. The government is assumed to invest as much of its budget in
flood defence projects as possible, assuming that it can meet the cost from its budget and reflecting
that the project requires a cost benefit ratio of 1:5 or higher (Environment Agency, 2009). The local
government creates a flood defence portfolio of potential flood defence projects it can select to fund.
These projects will focus on the highest risk properties which have no existing defences, and capture
as many houses surrounding this while meeting the above approval criteria.
As the focus of the research is on surface water flood risk flood defences are represented by
investment in SUDs. There is limited information on the costs and benefits of SUDS, particularly for
urban retrofit, with data often location and case specific. When considering new developments Defra
(2011) assumed that SUDS could reduce flood damage by up to 35%. Overall, evidence suggests that
SUDS are cheaper to build (up to 30%) than traditional drainage systems, addressing the concern of
Project 308438 • Risk scenarios and analysis: London case study 14
some developers that SUDS give rise to significant costs (Defra, 2011; Environment Agency, 2007).
This will be case specific, but even in complex cases costs are only assumed to be up to 5% higher
than traditional drainage system costs.
Defra (2011) report that on average 38% of minor developments and 58% of major developments are
now being built by developers with SUDS systems. It is not clear to what standards they are being
built to. It is assumed that 40% of build was accounted for by Minor Development with the remaining
60% accounted for by Major Development. This implies that 50% of build does not have SUDS.
Given the above data it is assumed that 50% of new builds created in the model will already have
SUDS at no extra cost. For retrofitting there is limited data and so it is assumed that the cost per house
to the government is £2000. In both cases risk will be reduced by 35% to estimate the potential
benefits of avoided damage. In future analysis we will also test the sensitivity of the above parameters
and options such as 1) that all new builds will include SUDS at no extra cost and reduce risk by 35%;
2) a current situation where 50% of new builds already have SUDS; and 3) a scenario where that there
is no uptake of SUDS and risk is increased by 35%.
2.5.3 Modelling the insurance market and Flood Re scheme
It is assumed that all home owners will have flood insurance initially, and will be required to pay an
insurance premium. In reality insurers can deny flood insurance to high risk properties, but in order to
gain insight into how the Flood Re scheme will address these high risk properties it is assumed
everyone can gain insurance, although at a given price. As the focus of the modelling is on the
technical rather than commercial risk the insurer will calculate flood risk based on the probability of a
given flood event and its associated damage (outlined in section 2.5.1), which will be used to calculate
the flood premium.
Home owners will also have an estimated insurance excess on their policy, which they will need to
cover in repairing damages following a flood event. It is assumed that the excess amount is non-
negotiable, and will be 1% (value based on expert opinion) of the insured value of the property.
Following any flood events the excess will also increase by 1/3rd
.
The insurer assets will reflect the income they receive in premiums and the compensation payments
they make following flood events, reflected in their loss ratio which will be calculated following each
flood event. Based on this value the insurer can increase or decrease prices accordingly.
Within the model there is also the option to ‘switch on’ the Flood Re scheme. The Flood Re scheme
will have assets in terms of a levy already paid as part of the insurance premium of households. If the
insurer decides the flood risk of a house is too high they are re-insured into the Flood Re scheme (if
built before 2009). The insurer will have to pay to re-insure the house into Flood Re, and income will
also be generated from the household premiums, capped based on the council tax band of the property
(where band A = £210; B = £210; C = $246; D = £276; E = £330; F = £408; G = $540; H = $540). In
this way the total compensation the insurer pays following a flood will be lower when the Flood Re
option is selected, as they are no longer required to compensate houses at highest risk.
Project 308438 • Risk scenarios and analysis: London case study 15
2.5.4 Modelling the London housing market
As part of the London case study a particular area of interest is the implication of flood risk,
un/insurability, and the role of Flood Re on house prices and the housing market as a whole. There is
no current consensus on the impact of flood risk and insurability on property prices. Some empirical
studies have shown that property values drop immediately after a flood event, although they can
recover over time as memory of the event fades. For example, in a study of Albany, USA, Atreya and
Ferreira (2014) do not find a significant discount for floodplain properties before a flood event, but in
the wake of the 1994 flooding a significant discount is seen in property prices, which then diminishes
over time. Similarly, the UK based study of Lamond and Proverbs (2006) suggests that the impact of
flooding on house prices is temporary, lasting less than three years (Lamond and Proverbs, 2006). In
contrast, a UK study based on data on repeat sales found that high flood risk had no effect on property
values in areas with no recent flood events (Lamond et al., 2009), suggesting it is the actual flood
occurrence rather than perceived flood risk which has the largest effect on valuation of property prices.
To investigate these issues for the London case study the ABM has been developed to incorporate a
representation of the London housing market. The initial set up is based on the spatial location of
residential buildings from the UK Buildings database, with properties assigned data on property type,
age, council tax band, average property value per house type, level of flood risk, and the flood depth
and associated damage, if any, given a 1 in 30, 1 in 100, or 1 in 200yr flood event.
Consumers form the basis of the housing market with the ability to own, buy or sell houses. The initial
conceptualisation and design of the housing market is based on that of Zhang et al., (2013), who
propose a double auction market aimed at achieving high levels of market efficiency quickly by
running rounds of trading between buyers and sellers. Their model provides three choices: The buyer
can bid an amount for a property; the seller can ask an amount for a property; and there will be a
transaction price for the property. Where the bid and ask cross, the transaction price will be equal to
the earlier of the two. So, for example, a homebuyer forms a bid price between his income level,
adjusted by an affordability multiplier, and zero. A home seller randomly forms an ask price between
the maximum reasonable market price and his buying price of the property. A selected home buyer
then compares his bid with the best ask. If his bid is above the best ask, he accepts the best ask and the
transaction occurs at the best ask. If his bid is below the best ask (or there is no best ask) and there is
no best bid, it becomes the best bid. If his bid is below the best ask (or there is no best ask) and above
the best bid, it overrides the best bid. If his bid is below the best bid, the bid is ignored. The rule also
applies to a selected home seller. If his ask is below the best bid, a trade occurs at the best bid.
As in the model of Zhang et al., (2013) houses can be sold by consumers and the Bank agent, who can
repossess properties and sell these at a discounted rate on the market. In an extension of this the
London ABM also models future residential developments (see section 2.5.4 below), and as such
houses are also sold by the Developer. In running the model it is assumed that there are three separate
markets for the person, developer and bank to sell houses. It is assumed that consumers prefer newer
houses and so the developer market is run first, then the person market, and then the bank market. It is
assumed home buyers will also aim for the most expensive house and best possible house type first
Project 308438 • Risk scenarios and analysis: London case study 16
and so houses are also sold based on this prioritisation from detached, semi-detached, and terraced,
down to flats.
Based on the modelled information on the number of houses on sale before the market run and houses
sold by the end of the run a ratio is created for each house type, and used to increase/decrease the
value of houses of this type accordingly. While house prices reflect the broader demand in the housing
market, consequences of flood events, insurability, and risk perception are incorporated as part of the
property owner’s decision making process when buying. Within the model buyers may or may not be
risk averse in their decision to consider past flood history and the flood risk level of a property. Risk
averse buyers can look back at a houses history, to a maximum of three years, to see whether it has
flooded (Lamond et al., 2009), and will not purchase a property if it has been flooded in the current
year. For houses flooded in the previous year 50% of risk averse buyers would not consider buying,
dropping to 25% for property flooded 2 years earlier. As such, there will be less demand for properties
where previous flood events have recently occurred.
In terms of selling decisions, property owners have three motives for selling their property in the
model. Firstly if they cannot afford the property fees i.e. annual fees such as the mortgage exceed their
income, in which case the bank can repossess their property. For flood affected home owners this will
also reflect the cost of insurance premiums, and the excess value they will have to cover following a
flood event. In the case of a property being uninsured this will reflect the total damage costs of the
flood which will need to be repaired. If a home owner cannot afford this additional cost then the house
is not repaired and the property value declines by this amount. Secondly a property owner may decide
to sell if they believe they can make a profit if the property value increases significantly above the
purchase value, and thirdly they can simply decide to move to a different house for other social factors
not considered explicitly within the model.
2.5.5 Modelling future development
As part of the housing market future building developments of residential properties are modelled.
This also allows for the role of developers and the local government to be investigated in terms of
testing different policy options and incentives for proposing and approving developments in high flood
risk areas.
In summary, the developer will establish the number of houses it wishes to build, based on the current
unmet demand for housing in the model. The developer will locate land for development based on
outlines of opportunity areas proposed for Greater London and the land value. Land value is calculated
based on the value of the surrounding houses. The developer will then create a development proposal
which will include the land it wishes to develop, the house type and the average value once built, and
the income cost ratio for the developer. If the income is higher than the expected cost then the
developer will submit this to the government for evaluation. In the model a development proposal will
be approved in 75% of the cases without a specific reason. Otherwise the development proposal will
be approved if the proposed flood risk of the development is lower than the maximum flood risk. If
this is also not the case the development proposal can still be approved if the development value at
risk (proposed building value if the house will be hit by a 1 in 200 year flood) divided by the
Project 308438 • Risk scenarios and analysis: London case study 17
development income (proposed land value + proposed tax fee) is lower than the development approval
ratio.
All new properties in the model will be assigned their property type, a council tax band, flood risk
level, insurance premium and excess value, and the flood depth and associated damage, if any, given a
1 in 30, 1 in 100, or 1 in 200yr flood event.
2.5.6 Using the ABM to inform MSPs
The ABM has been developed in such a way that it is able to address multiple questions to help inform
the MSPs in London. Key questions which will be addressed include:
How Flood Re could be used to enhance resilience both in terms of physical risk reduction and
financial resilience. For example:
o how will design and coverage of the scheme effect risk (e.g. including/excluding new
build properties)
o modelling the different options for transition from Flood Re to risk based pricing
o How will present risk and risk under future climate change scenarios.
How insurance affordability could affect property prices and household decisions to buy/sell
in certain areas.
o Will `Ghetto' areas develop and if so where?
The implications of changing stakeholder/consumer information and knowledge on flood
awareness e.g. via communication, flood insurance, and government level adaptation
strategies
The effects of incentives to install flood resistance and/or resilience measures
o which agents could be best positioned to incentivise such measures e.g. government,
insurer, and developer?
How much could Property Level Protection measures (PLPMs) reduce flood risk/economic
damage?
What are the benefits of SUDs and other adaptation measures?
How could government affect flood risk
o i.e. government agents can refuse, permit or promote development in particular
locations?
The results of this research will contribute to the overall assessment of the existing public-private
flood insurance partnership and the proposed new insurance scheme Flood Re, and how this could
influence London’s resilience to flood risks today and in the future. This work is ongoing and a full
description of the method, sensitivity analysis, and results from part (ii) of the London case study will
be included within Deliverable 7.4.
Project 308438 • Risk scenarios and analysis: London case study 18
3 Results
3.1 Surface water flood risk analysis
Extreme Value Analysis (EVA) was used to calculate the return levels of hourly precipitation, for each
return period, and for each of the 71 grid cells covering the Greater London study area. Results for the
grid cells were in the approximate range of 15-19.5mm, 17-23mm, and 18.5-25.5mm per hour for the
1 in 30, 100, and 200 year return periods respectively (Figure 5).
Figure 5: Return level of hourly precipitation (mm) for each return period across the 71 grid cells
The EVA provided the thresholds on which to estimate potential flood events of given return periods
and linked to flood depth and damage data. Using the Drain London flood depth maps the results
indicate that 312,551 to 392,758 properties could be at risk of surface water flooding (Table 2). This is
lower than the estimated 800,000 properties cited as being at risk from surface water flooding by the
GLA (2011). This may reflect the use of more accurate surface water flood depth maps from Drain
London, and the primary focus on residential properties only. Whilst there may also be some
limitations in the accuracy of the residential building database, the coverage was good with only
0.26% of total residential buildings being unclassified and as such excluded.
Table 2: Estimated economic impact of surface water flooding on residential buildings in Greater London
Return
period
Number of
properties affected
Damage to
building fabric
Damage to
building contents
Total damages
1/30yr 312,551 £2,319,104,560 £2,165,351,098 £4,484,455,658
1/100yr 352,942 £2,735,821,529 £2,498,683,701 £5,234,505,230
1/200yr 392,758 £3,140,705,414 £2,825,877,579 £5,966,582,992
Assuming a flood event affected the whole of Greater London results in total economic damages of
approximately £4.48bn to £5.97bn for 1/30 and 1/200yr flood events (Table 2), highlighting the
Project 308438 • Risk scenarios and analysis: London case study 19
potentially severe economic implications of large scale events in London. As a comparison flood
claims following the UK 2007 floods (including surface water flooding) were estimated to be £3.2bn.
Figure 6 provides a breakdown of economic costs per property type and return period. Whilst the
largest proportion of residential building type in London is flats it should be noted that the building
dataset represents the spatial footprints of residential buildings. In the case of flats this will not
represent the upper storeys and individual flat units. However, this is not considered an issue for the
analysis as it is assumed that only the ground floor properties would be affected by surface water
flooding.
Figure 6: Breakdown of total economic damage by property type and return period
Figure 7 presents maps of flood risk, aggregated at a ward level for Greater London. The maps
highlight the spatial pattern of risk based on the flood extent and depth data, but also represents the
concentration and number of residential buildings in each ward, and the type of properties affected.
The figure highlights particular risk hotspots in Greater London, particularly in the South and South-
East.
Project 308438 • Risk scenarios and analysis: London case study 20
Figure 7: Spatial maps of total economic damage to residential properties from surface water flooding for a) 1 in 30yr return
period, b) 1 in 100yr return period, and c) 1 in 200yr return period
However, surface water flooding would not affect the whole of Greater London at the same moment in
time. A benefit of using the probabilistic spatial WG is that long time series of daily events can be
provided for each grid cell (and aggregated to annual time-steps for use in the ABM as discussed
above), and as the WG is spatially coherent events affecting multiple grid cells can also be assessed,
with results reflecting both the intensity and spatial patterns of events across Greater London. Figure 8
presents a snap shot of a 100 year time slice showing the annual occurrence of flood events of
different probabilities for three grid cells in Greater London, for the baseline (top) and 2050 high
scenario (bottom). The graphs illustrate how areas of Greater London may be affected at different
times or simultaneously by surface water flood events, and how climate change is likely to increase
the severity and frequency of events in the future with the potential for areas to be affected in
successive years. Figure 9 illustrates the spatial extent and severity of two different rainfall events
identified.
Project 308438 • Risk scenarios and analysis: London case study 21
Figure 8: Annual time-series of flood events for three individual grid cells in Greater London
Figure 9: Spatial maps showing two examples of extreme precipitation events and severity
Project 308438 • Risk scenarios and analysis: London case study 22
Table 3 presents a summary of the estimated daily event frequency, average number of grid cells
affected and average damage per event and per year. The results are estimated based on the entire
hourly data array and presented for daily events which fall within the thresholds of individual return
periods.
The average values shown reflect the wider spatial extent of 1 in 30yr events, resulting in larger costs
per event than for 1 in 100 or 1 in 200yr events (although it should also be noted that average values
also mask large variation in flood losses of individual events). For example 1 in 30yr events affected
an average of 8 grid cells under the baseline scenario, compared to 5 grid cells for 1 in 200yr events.
Similarly, even though damages during 1 in 100 and 1 in 200 year flood events have the potential to
be much higher, the increased frequency of events of 1 in 30 years means that their contribution to
overall flood risk is very large (Table 3 and Figure 10).
Figure 10: Present and future flood risk (million £/yr) for Greater London
Project 308438 • Risk scenarios and analysis: London case study 23
Table 3: Frequency of surface water flood events and associated damages
Baseline 2030 High 2050 High
Return
Period
Total
number
of daily
events
Average
damage per
daily event
(£ 2010)
Average
annual
costs (£
2010)
Average
number
of grid
Cells
affected
per
daily
event
Total
number
of daily
events
Average
damage per
daily event
(£ 2010)
Average
annual
costs (£
2010)
Average
number
of grid
Cells
affected
per
daily
event
Total
number
of daily
events
Average
damage per
daily event
(£ 2010)
Average
annual
costs (£
2010)
Average
number
of grid
Cells
affected
per
daily
event
1/30yr 1329 489,944,906 593,021 8 2081 440,847,688 996,123 7 2193 529,741,051 1,058,035 9
1/100yr 434 384,744,087 152,076 5 806 386,906,864 284,014 6 855 397,030,988 309,164 6
1/200yr 275 404,253,759 101,248 5 562 442,284,054 226,379 6 609 463,169,045 256,894 6
All
RPs 1588 535,835,292 774,960 7 2450 522,359,015 1,165,754 7 2537 602,463,466 1,392,031 8
Project 308438 • Risk scenarios and analysis: London case study 24
The analysis highlights that overall event frequency could increase by 54% by the 2030s and by 60%
by the 2050s compared to the baseline period, although the average area of events remained largely
unchanged. The relatively small change in frequency of events between the 2030s and 2050s is in line
with other studies. For example, the study by Sanderson (2010) for London used UKCP09 data to
illustrate that rainfall events are likely to become more frequent in the future, particularly between the
present and 2040s. This could be because extreme rainfall events, particularly for longer return
periods, are already near to the maximum possible return levels. Hence, there will be a maximum
amount of precipitable water in the atmosphere and whilst the atmosphere can hold more water as it
warms, the incremental change in these extremes could become progressively smaller as the
atmosphere warms.
However, average annual damages were projected to increase by 50% and 80% by the 2030s and
2050s respectively, suggesting an increase in severity of events when they do occur. Based on the
daily event data, the expected annual damage (EAD) is calculated as £103, £184 and £198
million/year for the baseline, 2030 high and 2050 high climate change scenarios respectively.
An important component of the spatial WG is also the ability to provide probabilistic results. Table 4
presents results for all surface water flood events (i.e. where events are equal to and exceed the 1/30yr
return threshold), for the baseline and 2030 high and 2050 high climate scenarios, and also
disaggregated across the 100 model runs to provide an indication of the model uncertainty. Results are
presented for the median (50th percentile) and 10
th and 90
th percentile range.
Project 308438 • Risk scenarios and analysis: London case study 25
Table 4: Percentile results showing the frequency of surface water flood events and associated damages for All RPs
Baseline 2030 High 2050 High
frequency
average cost
per year
average cost
per event frequency
average cost
per year
average cost
per event frequency
average cost
per year
average cost
per event
10th 13 90,549,441 175,145,317 14 155,401,104 244,841,515 13 180,389,736 306,969,764
50th 17 289,261,556 526,178,194 28 457,602,303 498,244,867 25 562,214,261 583,049,021
90th 23 585,393,662 914,098,752 57 1,117,925,503 853,175,802 48 1,172,357,582 1,073,215,947
Project 308438 • Risk scenarios and analysis: London case study 26
The results suggest that the frequency of events will increase by the 2030s and 2050s compared to the
baseline, however, with a larger increase by the 2030s compared to the 2050s. In contrast the average
cost of surface water flood events per year increases from approximately £289 million per year (50th
percentile) in the baseline, to £458 million and £562 million per year in the 2030s and 2050s
respectively. As indicated above this suggests that while the frequency of events may increase in the
shorter term this may include more smaller scale events, compared to the longer term results which
have less events but when they do occur they are of a larger spatial extent and severity. This is also
highlighted when looking at the average cost per event. This declines from approximately £526
million per event (50th percentile) in the baseline to £498 million in the 2030 high scenario, but
increases to £583 million under the 2050 high scenario
Figures 11 and 12 present these results, along with the 10th and 90
th percentile range. In both examples
the range in results increases from the baseline period to the future time periods, representing the
climate model uncertainty underlying the precipitation projections.
Figure 11: Cost of surface water flood events per year. Results are presented for the median and 10th and 90th percentile
range
Figure 12: Frequency of surface water flood events per year. Results are presented for the median and 10th and 90th
percentile range
Project 308438 • Risk scenarios and analysis: London case study 27
4 Discussion of Results
4.1 Surface water flood risk analysis
The main aim of the above work was to establish a link between outputs from a probabilistic urban
spatial WG and the detailed flood depth maps from Drain London to evaluate current and future risk
of surface water flooding in Greater London. Comparing literature on fluvial or coastal flood risk and
surface water flood risk highlights that the latter has received much less attention. Concerns have been
raised that pluvial flood risk across Europe may not be fully recognised, with some member states
giving a much lower priority to this type of risk (European Water Association, 2009). In the UK the
recent Climate Change Risk Assessment, which assessed changing flood risk under future climate
change, also focused on tidal and fluvial flooding, whilst implications of climate change for pluvial
and surface water flooding were not considered (although evidence from studies conducted by e.g. the
Environment Agency, Defra, and the GLA were considered) (Adaptation Sub-Committee, 2012).
Correspondingly, there has been less emphasis on quantitative assessments of economic costs of
surface water flooding for the present day and under future climate change, with a need to improve
methods and models for loss estimation. As such the above methodological framework applied to
Greater London provides a step to address this current gap in research.
It was estimated that 312,551 and 392,758 residential properties would be at risk of surface water
flooding following a 1 in 30 and 1 in 200 year event in Greater London. This was lower than the
estimated 800,000 properties cited as being at risk by the GLA (2011). However, it was postulated that
this could reflect the use of a different methodological approach, including the more recent and
detailed Drain London depth maps, and use of the UK Buildings dataset which carries its own
limitations. As further information on the method underlying the figure cited by the GLA could not be
found it is hard to provide a further comparison and fully verify this difference at this stage. Whilst
results may be considered conservative the economic damages estimated were still significant, ranging
from £4.4bn to £5.97bn for 1 in 30 and 1 in 200 year events across Greater London. The spatial
analysis also allowed risk hot spots to be identified (Figure 4), namely wards in the south and south-
east, where flood damage could reach approximately £50 million.
Given the localised nature of surface water flood events the analysis also allowed specific daily events
to be identified. Based on this event dataset the analysis highlights that overall event frequency could
increase by 54% by the 2030s and by 60% by the 2050s compared to the baseline period, although the
average area of events remained largely unchanged. Average annual damages were projected to
increase by 50% and 80% by the 2030s and 2050s respectively, suggesting an increase in severity of
events when they do occur. In comparison, Defra estimate flood damage from surface water run-off
could increase by 60-220% over the next 50 years linked to changing precipitation patterns due to
climate change, and urbanisation (Adaptation Sub-Committee, 2012). Again, results of this study
appear to be conservative, falling with in the lower end of this estimate. This discrepancy may be due
to the inclusion of other damage costs within the Defra estimates, but also that this study does not
consider additional effects of urbanisation such as new developments, especially in flood areas of high
flood risk, and increasing manmade surfaces and reduced greenspace.
Project 308438 • Risk scenarios and analysis: London case study 28
The probabilistic analysis also highlighted how annual damages were projected to increase, although
the frequency of events was higher under the 2030 high scenario compared to the 2050 high scenario,
suggesting an increase in severity of events when they do occur. Results were presented for the 50th,
10th and 90
th percentile range and indicated the uncertainty in results due to underlying uncertainty in
the climate model projections. This uncertainty increases for both the 2030 and 2050 high scenarios
compared to the baseline.
Based on the daily event data, the expected annual damage (EAD) is calculated as £103, £184 and
£198 million/year for the baseline, 2030 high and 2050 high climate change scenarios respectively.
Assuming all else remains static this highlights both the economic impact that climate change could
have, but also the potential cost effectiveness of benefits that could be gained through investment in
adaptation options such as property level protection measures, SUDs, and urban greening. The role of
adaptation for risk reduction will be investigated further in the next stage of the analysis via the ABM
(see section 2.5.2). Additionally the methodology presented here could also be used to assess benefits
of adaptation. While there is limited data on the specific benefits of such measures on resultant flood
depth and damage that could be applied to Greater London as a whole, one option would be to conduct
a sensitivity analysis of the perceived benefits of adaptation. For example, assuming an upgrade to the
drainage system in which it could cope with a 1 in 100 year event rather than the current 1 in 30 year
event.
Whilst the methodology provides a novel approach to estimating surface water flood risk for a large
area there are also issues and limitations of any such study. In particular some important points to be
noted are:
Flood management is increasingly adopting a risk based approach, often using a range of
flood risk probabilities to provide a range of costs. However, as noted by Ward et al., (2011)
there is little insight into how these return periods are selected, how many should be
considered, and their impact on the outcomes of the study. This study adopted three return
periods in line with the available Drain London flood depth maps and the methodological
approach of the GLA.
Whilst the study benefits from the use of the spatial WG, which considers geographic and
topographic variations, other important processes which can affect rainfall, and which occur
over small spatial scales, are excluded and could result in model uncertainty.
Although considerable effort has been made to improve the rainfall model care should be
taken in using the WG to reproduce and analyse extreme precipitation events. The user is
advised to carry out uncertainty analysis based on the WG runs (Jones et al., 2009), and it is
noted that uncertainty will increase for longer return periods. The version of the WG used in
this study allows the effect of climate model uncertainty in projected future impacts to be
explored as discussed above.
As noted the probabilistic spatial WG allows uncertainty related to the physical climate
properties to be investigated. However, uncertainty surrounding the economic damages is
harder to ascertain. Flood depth-damage functions have limitations due to uncertainty in the
underlying data and assumptions which preclude their creation, and the influence of other
factors than depth on flood damage, such as velocity, duration, rise rate, and time of
occurrence (Messner et al., 2007). Even so, water depth is considered the single most
Project 308438 • Risk scenarios and analysis: London case study 29
important factor for building damage (Elmer et al., 2010; Messner et al., 2007) and so is a
good characteristics to use.
The accuracy of such estimates can be difficult to validate, although comparisons can be made
to historical events and damage data. This is also reflective of the flood depth-damage
functions used by this study which are generally comprised of synthetic data and not directly
derived from analysis of historical flood events (Penning-Rowsell et al., 2010).
The depth damage functions are also based on national average data and so may underestimate
losses for London.
Additionally, it is reported that flood incidents can be accompanied by significant emergency
costs such as additional expenditure due to increased demand for police, fire and ambulance
services. It has been estimated that these are equivalent to 10.7% of property damages
(Penning-Rowsell et al., 2010). The above estimates reflect direct costs to residential property
only, but total costs including indirect effects to the wider region and economy, could be
substantially higher (e.g. see Crawford-Brown et al., 2012 for further discussion of indirect
losses).
Likewise, flood events can result in environmental and social impacts which are not
considered here e.g. environmental and ecological impacts of degraded water quality, and
psychological effects caused by displacement and stress following a flood event.
In the study affected houses were established by overlying the flood depth maps onto the
spatial footprints of residential buildings. This provides a simple interpretation of the houses
which would be affected given the limited data on the specific architecture of buildings (e.g.
height above pavement, whether they have basements etc.). This was a consequence of
developing a methodology which could be applied to a large region rather than focusing on
specific properties and streets.
Finally, the study considers one of the main drivers for flood risk – precipitation (and the
impact of climate change on precipitation regimes). However, other drivers such as
urbanisation will also be important. The impact of residential developments in areas of high
flood risk will be an area investigated further in the second stage of the analysis via the ABM.
Project 308438 • Risk scenarios and analysis: London case study 30
5 Conclusions
The report presents a framework for evaluating current and future risk of surface water flooding in
Greater London. The case study highlights how this risk is expected to change in the future under
projections of climate change. Daily event frequency could increase by 54% by the 2030s and by 60%
by the 2050s compared to the baseline period. Average annual damage to residential properties is
projected to increase by 50% and 80% by the 2030s and 2050s respectively, suggesting an increase in
severity of events when they do occur. Based on the daily event data, the expected annual damage
(EAD) is calculated as £103, £184 and £198 million/year for the baseline, 2030 high and 2050 high
climate change scenarios respectively.
The risk analysis provides a useful tool for decision making and risk management, and through the
proposed incorporation of this data in the ABM aims to support and justify future adaptation
strategies. These will be particularly important for Greater London where surface water flooding is
considered to be the most likely cause of flood events, and probably the greatest short-term climate
risk.
As previously discussed a particularly interesting aspect of flood management in England is the
public-private partnership on flood insurance between the UK government and the insurance industry
and the new flood insurance system Flood Re. The proposed system, which creates an insurance pool
for properties at high risk of flooding, is presented by government and industry as a roadmap to future
affordability and availability of flood insurance, with an anticipated run-time of 20 to 25 years (Defra
and ABI 2013). Flood Re was proposed by government in summer 2013, with the political debate
about the system and its mechanisms continuing. The current aim is to finalise and implement the new
scheme by mid-2015, and as such the analysis of the scheme through the ABM proposed here is
highly relevant for informing the MSP.
The future link to the ABM will allow an estimate of costs and benefits of adaptation strategies, and
their role in risk reduction, to be investigated. In addition, ongoing discussions with key stakeholders
have highlighted specific questions and issues of interest. Through the ABM the following issues can
be investigated:
The resilience of the new Flood Re scheme to increases in surface water flood risks
How the design and coverage of the scheme can affect flood risk. For example how will the
inclusion/exclusion of post 2009 properties impact on the Flood Re scheme
How Flood Re could be used to enhance resilience. For example including clauses within the
scheme whereby homeowners need to demonstrate resilient repairs or their excess increases or they
are potentially excluded from Flood Re
Issues of affordability and availability of insurance at the end of the 25-year transition period
that Flood Re is supposed to operate for.
Modelling the transition from Flood Re to risk based pricing and mechanisms for this. For example, a
gradual removal of properties over time, staggered removal of properties based on council tax bands, or
a gradual increase in the price of premiums in Flood Re
Project 308438 • Risk scenarios and analysis: London case study 31
The above issues and questions are all highly relevant aspects for the ongoing regulatory and political
approval process for Flood Re, which have until now not received sufficient attention due to lack of
data or analysis. Our work addresses this gap and our findings are expected to provide important input
to the current discussion about the design and operation of Flood Re, particularly with regards to
incentivising flood risk reduction measures.
Project 308438 • Risk scenarios and analysis: London case study 32
6 References
Adaptation Sub-Committee, (2012) Climate Change - Is the UK preparing for flooding and water
scarcity? Committee on Climate Change, London, p. 102.
Atreya, A., Ferreira, S. (2014) Seeing is Believing? Evidence from Property Prices in Inundated
Areas. Risk Analysis, n/a-n/a.
Bevan, K., Hall, J., (2014) Applied Uncertainty Analysis for Flood Risk Management. Imperial
College Press, London, p. 684.
Brouwers, L., Boman, M., (2011) A computational agent model of flood management strategies, in: do
Prado, H.A., Luiz, A.J.B., Filho, H.C. (Eds.), Computational Methods for Agricultural Research:
Advances and Applications. IGI Global, Hershey, PA, USA, pp. 296–307.
Coulthard, T.J., Ramirez, J., Fowler, H.J., Glenis, V. (2012) Using the UKCP09 probabilistic
scenarios to model the amplified impact of climate change on river basin sediment yield. Hydrol.
Earth Syst. Sci. Discuss. 9, 8799-8840.
Crawford-Brown, D., Syddall, M., Guan, D., Li, J., Jenkins, K., Hall, J., Beaven, R., Thoung, C.,
Anger-Kraavi, A. (2012) Exploring the Vulnerability of London’s Economy to Climate Change
Impacts. Environment and Planning A Submitted.
Dawson, R.J., Speight, L., Hall, J.W., Djordjevi, S., Savic, D., Leandro, J. (2008) Attribution of flood
risk in urban areas. Journal of Hydroinformatics 10.4, 275-288.
Defra, (2008a) Consultation on policy options for promoting property-level flood protection and
resilience, in: Department for Environment, F.a.R.A. (Ed.). Department for Environment, Food and
Rural Affairs, London, p. 64.
Defra, (2008b) Resilience grants pilot projects, in: Department for Environment, F.a.R.A. (Ed.).
Department for Environment, Food and Rural Affairs, London, p. 15.
Defra, (2011) Commencement of the Flood and Water Management Act 2010, Schedule 3 for
Sustainable Drainage: Impact Assessment. Defra, London, p. 42.
Defra and ABI (2013), Flood Re Proposal: Memorandum of Understanding. London:
Department for Environment, Food and Rural Affairs, and the Association of British Insurers.
Available at:
https://consult.defra.gov.uk/flooding/floodinsurance/supporting_documents/20130626%20Fl
ood%20Insurance%20MOU%20June%202013%20unprotected.pdf [Accessed 8 July 2013].
Ellis, J.B., Viavattene, C. (2014) Sustainable Urban Drainage System Modeling for managing Urban
surface water Flood Risk. Clean - Soil, Air, Water 42, 153-159.
Elmer, F., Thieken, A.H., Pech, I., Kreibich, H. (2010) Influence of flood frequency on residential
building losses. Nat. Hazards Earth Syst. Sci., 10, 2145-2159.
Environment Agency, (2007) Science Report – A review of the cost-benefit of undertaking SUDS
retrofit in urban areas. Environment Agency, Bristol, p. 33.
Project 308438 • Risk scenarios and analysis: London case study 33
Environment Agency, (2009) Flooding in England:A National Assessment of Flood Risk, in: Agency,
E. (Ed.). Environment Agency, Bristol, p. 36.
Environment Agency, (2010) Case study – Surface water flooding in London, July 2007. Environment
Agency, Avaliable at: http://www.environment-
agency.gov.uk/research/library/publications/115643.aspx.
European Water Association, (2009) EWA Expert Meeting on Pluvial Flooding in Europe, Final
Report. EWA, Brussels, Belgium.
Falconer, R.H., Cobby, D., Smyth, P., Astle, G., Dent, J., Golding, B. (2009) Pluvial flooding: new
approaches in flood warning, mapping and risk management. Journal of Flood Risk Management 2,
198-208.
Fowler, H.J., Ekström, M. (2009) Multi-model ensemble estimates of climate change impacts on UK
seasonal precipitation extremes. International journal of climatology 29, 385-416.
GLA, (2011) The London Plan: Spatial Development Strategy for Greater London. Greater London
Authority, London.
Greater London Authority, (2009) London Regional Flood Risk Appraisal, in: Plan, T.M.L. (Ed.).
GLA, London, p. 62.
Greater London Authority, (2011) The London Climate Change Adaptation Strategy. Greater London
Authority, London, p. 125.
Hall, J., Sayers, P., Dawson, R. (2005) National-scale assessment of current and future flood risk in
England and Wales. Natural Hazards 36, 147-164.
Hall, J.W., Sayers, P.B., Walkden, M.J.A., Panzeri, M. (2006) Impacts of climate change on coastal
flood risk in England and Wales: 2030–2100. Philosophical Transactions of the Royal Society A:
Mathematical, Physical and Engineering Sciences 364, 1027-1049.
Harries, T. (2008) Feeling secure or being secure? Why it can seem better not to protect yourself
against a natural hazard. Health, risk and society 10, 479-490.
Harries, T. (2012) The anticipated emotional consequences of adaptive behaviour - impacts on the
take-up of household flood-protection measures. Environment and Planning A 44, 649-668.
HR Wallingford, (2012) Development of spatial indicators to monitor changes in exposure and
vulnerability to flooding and the uptake of adaptation actions to manage flood risk in England: Results
2012. HR Wallingford Ltd, Oxfordshire, p. 56.
IPCC, (2013) Summary for Policymakers. In: Climate Change 2013: The Physical Science Basis.
Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on
Climate Change in: Stocker, T.F., D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A.
Nauels, Y. Xia, V. Bex and P.M. Midgley. (Ed.). Cambridge University Press, Cambridge, United
Kingdom and New York, NY, USA., p. 29.
Project 308438 • Risk scenarios and analysis: London case study 34
Jones, P.D., Kilsby, C.G., Harpham, C., Glenis, V., Burton, A., (2009) UK Climate Projections
science report: Projections of future daily climate for the UK from the Weather Generator, University
of Newcastle, UK.
Kilsby, C., Jones, P., Harpham, C., Glenis, V., Burton, A., (2011) Spatial Urban weather Generator for
Future Climates, ARCADIA Task 3 Report. Avaliable at: http://www.arcc-cn.org.uk/wp-
content/pdfs/ARCADIA-7.pdf, p. 8.
Lamond, J., Proverbs, D. (2006) Does the price impact of flooding fade away? Structural Survey 24,
363-377.
Lamond, J., Proverbs, D., Hammond, F., (2009) Flooding and Property Values, in: Brown, S. (Ed.),
Findings in Built and Rural Environments (FiBRE). Royal Institution of Chartered Surveyors, London.
Maraun, D., Osborn, T.J., Gillett, N.P. (2008) United Kingdom daily precipitation intensity: improved
early data, error estimates and an update from 2000 to 2006. International journal of climatology 28,
833-842.
Messner, F., Penning-Rowsell, E., Green, C., Meyer, V., Tunstall, S., van der Veen, A., (2007)
Evaluating flood damages: guidance and recommendations on principles and methods. HR
Wallingford, UK p. 189.
Murphy, J.M., Sexton, D.M.H., Jenkins, G.J., Boorman, P.M., Booth, B.B.B., Brown, C.C., Clark,
R.T., Collins, M., Harris, G.R., Kendon, E.J., Betts, R.A., Brown, S.J., Howard, T.P., Humphrey,
K.A., McCarthy, M.P., McDonald, R.E., Stephens, A., Wallace, C., Warren, R., Wilby, R., Wood,
R.A., (2009) UK Climate Projections Science Report: Climate change projections, Met Office Hadley
Centre, Exeter, UK, p. 190.
ONS, (2001) Census: Standard Area Statistics (England and Wales), ESRC/JISC Census Programme,
Census Dissemination Unit. Mimas, University of Manchester.
Penning-Rowsell, E., Viavattene, C., Pardoe, J., Chatterton, J., Parker, D., Morris, J., (2010) The
Benefits of Flood and Coastal Risk Management: A Handbook of Assessment Techniques. Flood
Hazard Research Centre, London, p. 98.
Ramsbottom, D., Sayers, P., Panzeri, M., (2012).Climate Change Risk Assessment for the Floods and
Coastal Erosion Sector. UK Climate Change Risk Assessment. Defra, London.
Sanderson, M., (2010) Changes in the frequency of extreme rainfall events for selected towns and
cities, Report for Ofwat. Met Office, Exeter, p. 38.
Thurston, N., Finlinson, B., Breakspear, R., Williams, N., Shaw, J., Chatterton, J., (2008) Developing
the evidence base for flood resilience, FD2607/TR. Defra, London.
Ulbinaitė, A., Kučinskienė, M., Le Moullec, Y. (2011) Conceptualising and Simulating Insurance
Consumer Behaviour: an Agent-Based-Model Approach. International Journal of Modeling and
Optimization 1, 250-257.
Ulbinaitė, A., Le Moullec, Y. (2010) Towards an ABM-based framework for investigating consumer
behaviour in the insurance industry. ekonomika 89, 95-110.
Project 308438 • Risk scenarios and analysis: London case study 35
Ward, P.J., de Moel, H., Aerts, J.C.J.H. (2011) How are flood risk estimates affected by the choice of
return-periods? Nat. Hazards Earth Syst. Sci. 11, 3181-3195.
Zhang, H., Putra, H.C., Andrews, C.J., (2013) Modeling Real Estate Market Responses to Climate
Change in The Coastal Zone, The Computational Social Science Society of the Americas, Santa Fe,
USA, p. 18.